Raman spectroscopy method

Abstract

The invention relates to a method of Raman spectroscopy. The method comprises the following steps: determining a measurement spectrum by using a Raman spectrometer; determining a second derivative of the difference spectral function corresponding to a difference between the measured spectrum and a product of the scaling coefficient and a reference spectrum of contributions of interference effects contained in the measured spectrum; determining a coefficient value that minimizes a scaling coefficient of an error function, the error function comprising a term corresponding to an error of the difference spectral function due to peaks contained in the reference spectrum, the term comprising a sum of areas surrounded under the second derivative in all spectral regions in which the second derivative is positive, and / or wherein the second derivative is the sum of the areas surrounded below the second derivative in all spectral regions with negative; and determining a Raman difference spectrum corresponding to a difference between the measured spectrum and a product of the coefficient value and the reference spectrum.

Description

Technical Field

[0001] The present invention relates to a method of Raman spectroscopy, in particular to a method for eliminating the contribution of interfering influences contained in a measurement spectrum. Background Art

[0002] Raman spectroscopy is a technique for measuring the wavelength and intensity of inelastically scattered radiation (e.g., light) from a sample, thereby revealing the sample's chemical and structural makeup. Raman spectroscopy is based on the principle that monochromatic excitation light will be reflected, absorbed, or scattered depending on the specific molecule (e.g., protein, peptide, carbohydrate, cytokine, salt, etc.) receiving the incident radiation. Most of the energy is scattered at the same wavelength as the excitation light, known as elastic scattering or Rayleigh scattering. A much smaller amount (e.g., ~0.001%) is scattered (i.e., shifted) at a different wavelength, known as inelastic scattering or Raman scattering, whose wavelength depends on the molecular makeup of the sampled area. In Raman spectroscopy (interchangeably, Raman analysis and Raman spectroscopy), these wavelength shifts are captured in a Raman spectrum that includes the Raman scattered light (i.e., the Raman signal), which is analyzed to determine sample properties, both chemical and physical.

[0003] Raman spectrometers are currently used in a large variety of applications, including industrial and laboratory applications, to determine and provide measurement results of various measured variables of a medium. As an example, Raman spectrometers are used to determine the concentration of components contained in a medium, the pH value of the medium, the melt index of the medium and / or the cell motility of the medium.

[0004] A Raman spectrometer generally comprises a light source for emitting monochromatic excitation light to a sample of a medium, and a spectrometric unit for receiving Raman scattered light emitted from the sample and determining and providing a measurement spectrum corresponding to the intensity spectrum of the Raman scattered light. For example, the measurement spectrum is provided to an evaluation unit, which determines and provides a measurement result based on a previously determined model for determining a measurement result of a corresponding measured variable based on the spectral intensity of the measurement spectrum.

[0005] The model used in Raman spectroscopy to determine the measurement results of the measured variable can be determined, for example, based on measured spectra of samples exhibiting known values ​​of the corresponding measured variable and a detailed mathematical analysis of the sample spectra.

[0006] However, one problem associated with Raman spectroscopy measurements is that at least some of the information about the measured variable(s) contained in the measured spectrum of the medium may be obscured by the contribution of interfering effects contained in the measured spectrum. As an example, with respect to the determination of the concentration of the target analyte, the measurement of the concentration of the target analyte contained in a medium comprising a plurality of components may be impaired by the contribution of the interfering analytes to the spectral intensity of the measured spectrum in the spectral region of interest. With respect to a medium comprising a solute dissolved in a solvent, the measurement of the measured variable of the solute may be impaired by the contribution of the solvent to the spectral intensity of the measured spectrum.

[0007] As another example, spectral information used to determine a measured variable may be obscured by contributions from light contamination contained in the light received by the spectrometric cell of a Raman spectrometer. Depending on the Raman spectrometer used and / or the application in which it is employed, the light contamination may include, for example, fluorescence emitted by fluorescent materials located along the optical path extending from the light source to the spectrometric cell, the fluorescent materials being illuminated by the excitation light.

[0008] In many applications, the contribution of interfering influences contained in the measurement spectrum makes it difficult or even impossible to accurately determine the measured variable(s). Therefore, it is necessary to eliminate the contribution of interfering influences contained in the measurement spectrum.

[0009] One way to achieve this goal is to determine the Raman difference spectrum by subtracting a reference spectrum of the contribution of the interfering influence from the measured spectrum. As an example, the reference spectrum of the contribution of the interfering analyte contained in the medium can be subtracted from the measured spectrum of the medium. In order to eliminate the contribution of the interfering influence, the reference spectrum must be scaled according to the amplitude of the contribution of the interfering influence contained in the measured spectrum in a manner that takes into account the intensity ratio of the Raman spectrometer that provides the measured spectrum. For example, scaling is performed by determining the spectral intensity value of the Raman difference spectrum as the difference between the spectral intensity value of the measured spectrum and the product of the spectral intensity value of the reference spectrum on the corresponding optical spectral line and the scaling factor.

[0010] However, determining coefficient values ​​for the scaling coefficients that correspond to the magnitude of the contribution of the interfering effects contained in the measured spectra can be a daunting task, particularly in applications where the concentration of the interfering analyte is unknown or variable and / or the intensity of light pollution may vary. Under-subtracting by applying too small coefficient values ​​will result in Raman difference spectra in which the spectral information used to determine the measured variable(s) may still be obscured by the remaining contributions of the interfering effects. Over-subtracting by applying too large coefficient values ​​may alter or even remove spectral information that could otherwise be used to determine the measured variable(s).

[0011] Although experimental and / or manual determination of coefficient values ​​can be tedious and prone to error, some progress has been made in automating spectral subtraction.

[0012] The paper "Second-Derivative Variance Minimization Method for Automated Spectral Subtraction" published by Yvette L. Loethen et al. in Applied Spectroscopy, Vol. 58, No. 3, 2004, describes a second-derivative minimization (SDVM) procedure for automatically extracting the spectrum of a diluted component contained in a medium from a measured spectrum of a medium dominated by a main component (e.g., a solvent). According to this method, the value of the scaling factor of the Raman difference spectrum is determined by subtracting the corresponding scaled reference spectrum of the solvent from the measured spectrum, so that it minimizes the error function given by the variance of the second derivative of the smoothed Raman difference spectrum.

[0013] As outlined in the paper, the method offers the advantage that it is less susceptible to noise and fluorescence interference than other methods. However, its use is limited to applications where the variance of the second derivative of the smoothed Raman difference spectrum exhibits a minimum at an optimal value of the scaling factor that is sufficiently distinct to allow the value of the scaling factor that minimizes the variance to be determined with sufficient accuracy.

[0014] This aspect makes the method unsuitable for applications in which the variance is a shallow function of the scaling factor. The latter is for example the case for applications in which the solute and solvent spectra are highly correlated and / or comprise overlapping spectral peaks. The shallower the error function, the greater the error associated with the determination of the scaling factor value of the minimization function. Depending on the degree of correlation and / or spectral overlap, the error function may be so shallow that the error associated with the determination of the minimum exceeds the maximum allowed tolerance.

[0015] The paper "Asymmetric least-square baseline algorithm with peak screening for automatic processing of the Raman spectra" by Vitaly I. Korepanov, published in the Journal of Raman Spectroscopy, 2020; 1-5 (https: / / doi.org / 10.1002 / jrs.5952), discloses a method for determining a smooth baseline representing the contribution of interfering effects contained in the measured Raman spectrum by minimizing an error function given by the sum of two terms. The first term is an asymmetric weighted sum of the squares of the differences between the spectral intensity values ​​of the measured spectrum and the baseline, wherein the weighting factor applied to the differences due to the spectral intensity values ​​of the measured spectrum exceeding the corresponding spectral intensity values ​​of the baseline is significantly lower than the weighting factor applied to the differences due to the spectral intensity values ​​of the measured spectrum being less than the corresponding spectral intensity values ​​of the baseline. The second term is given by the product of a constant penalty factor and the sum of the squares of the second-order derivatives of the baseline at the individual optical spectral lines.

[0016] As recognized by the paper, in the peak region of the measured spectrum, the difference between the spectral intensity values ​​of the measured spectrum and the baseline may become large enough to cause an undesirable inflection point in the estimated baseline. According to the paper, this problem can be solved by performing an iterative process of determining weighting factors in a manner that takes into account the peak region.

[0017] Asymmetric least squares algorithms (ALS), such as those described in the above-cited paper, can also be used to determine the values ​​of the coefficients for the scaling factors used to determine the Raman difference spectrum by subtracting a scaled reference spectrum with contributions from interfering effects from the measured spectrum. In this context, they provide good results with respect to eliminating contributions from interfering effects (such as broadband luminescence light), showing a fairly smooth intensity profile.

[0018] However, a disadvantage of the corresponding ALS algorithms is that they penalize the second derivative of the reference spectrum. This essentially increases the penalty applied to any peak region that the reference spectrum may exhibit. As a result, the scaling factor values ​​determined based on the ALS algorithm for reference spectra that exhibit different peaks are often too small. This results in an insufficient subtraction, resulting in Raman difference spectra where the valuable spectral information may still be obscured by the contribution of interfering effects left in the Raman difference spectrum.

[0019] Therefore, there is still a need for further contributions in this area of ​​technology. As an example, there is a need for a method that can better eliminate the contribution of interfering effects contained in a measurement spectrum in an application, wherein the contribution exhibits a multi-peak reference spectrum and / or a reference spectrum that includes spectral features that are correlated and / or at least partially overlap with other spectral features contained in the measurement spectrum. Summary of the invention

[0020] The present disclosure includes a method of Raman spectroscopy, comprising determining and providing a Raman difference spectrum for performing Raman spectroscopic measurements of at least one measured variable of a medium in at least one application, wherein the measured spectrum of the medium includes contributions of interfering effects that at least partially obscure spectral features associated with the measured variable(s), the method comprising performing the following method steps in at least one application:

[0021] Determining at least one measurement spectrum of the medium using a Raman spectrometer;

[0022] providing a reference spectrum that accounts for the contribution of interfering effects contained in the measured spectrum;

[0023] determining a difference spectrum function corresponding to a difference between a measured spectrum and a product of a scaling factor and a reference spectrum;

[0024] determining the second derivative of the difference spectral function with respect to the spectral line;

[0025] Determining a coefficient value of a scaling factor corresponding to the magnitude of the contribution of the interfering effect contained in the measured spectrum such that the coefficient value minimizes an error function representing an error of the difference spectrum function as a function of the scaling factor, wherein the error function comprises a first term corresponding to an error exhibited by the difference spectrum function due to a peak contained in the reference spectrum, the first term comprising at least one of the following:

[0026] a first sum of the areas enclosed under the second derivative of the difference spectral function in all spectral regions where the second derivative of the difference spectral function is positive; and

[0027] the second sum of the areas enclosed under the second derivative of the difference spectral function in all spectral regions where the second derivative of the difference spectral function is negative; and

[0028] For at least one measurement spectrum determined by the Raman spectrometer, a Raman difference spectrum corresponding to the difference between the respective measurement spectrum and the product of the coefficient value and the reference spectrum is determined and provided.

[0029] The method provides the advantage that, in applications in which the reference spectrum is a multi-peak spectrum containing at least one peak, it enables accurate determination of coefficient values ​​corresponding to the magnitude of the contribution of interfering influences contained in the measured spectrum. In the context of a multi-peak reference spectrum, the first term contained in the error function provides the advantage that it is a V-shaped function of the scaling factor, which results in the error function exhibiting a clear and therefore accurately determinable minimum at the coefficient value corresponding to the magnitude of the contribution of the interfering influences contained in the measured spectrum. The accurate determination of the coefficient value provides the advantage that the Raman difference spectrum provided by the method is of high quality due to the elimination of the contribution of interfering influences achieved by adopting the coefficient value of the accurately determined scaling factor.

[0030] In some embodiments, the method includes filtering the measured spectrum and the reference spectrum using a smoothing filter, a second-order or higher-order smoothing filter, a Savitzky-Golay filter, a second-order or higher-order Savitzky-Golay filter, or a wavelet decomposition filter, and determining a difference spectral function and / or a second-order derivative of the difference spectral function based on the filtered measured spectrum and the filtered reference spectrum.

[0031] In some embodiments, the first term is determined as the sum or weighted sum of the first sum and the second sum, the integral or trapezoidal integral of the absolute value of the spectral value of the second-order derivative of the difference spectral function over the spectral range, the term given by the sum of the product of the first sum, the penalty factor and the second sum, or the term given by the sum of two products, wherein the first product is determined by multiplying the first sum by the penalty factor, and the second product is determined by multiplying the difference between one and the penalty factor by the second sum. In addition or as an alternative, the first sum is determined as the integral or trapezoidal integral of the rectified linear unit of the spectral value of the second-order derivative of the difference spectral function over the spectral range, and / or the second sum is determined as the integral or trapezoidal integral of the rectified linear unit of the negative one times the spectral value of the second-order derivative over the spectral range.

[0032] According to an embodiment, the error function further includes a second term, and the second term includes at least one of the following:

[0033] corresponds to an error in the difference spectral function due to negative spectral intensity values ​​of the difference spectral function caused by over-subtraction;

[0034] is given by the sum of the areas enclosed beneath the difference spectral function in all spectral regions where the spectral intensity values ​​of the difference spectral function are negative; and

[0035] It is determined as the integral or trapezoidal integral of the rectified linear unit over the spectral range of minus one times the spectral intensity value of the difference spectral function.

[0036] According to an improvement of this embodiment, the error function further includes a third term, and the third term includes at least one of the following:

[0037] corresponds to the error in the difference spectrum function due to insufficient subtraction;

[0038] is given by the sum of the areas enclosed beneath the difference spectral function in all spectral regions where the spectral intensity values ​​of the difference spectral function are positive; and

[0039] The rectified linear unit integral or trapezoidal integral of the spectral intensity values ​​as a function of the difference spectrum is determined over the spectral range.

[0040] In some improved embodiments, the error function is given by:

[0041] the sum or weighted sum of the first, second and third terms; or

[0042] The first term, the product of the penalty factor and the second term, and the sum of the third term.

[0043] In certain embodiments according to the improved embodiment, in each application, whether the reference spectrum is a multi-peak spectrum or a smooth spectrum, the coefficient value is determined based on an error function including a first term, a second term, and a third term.

[0044] Further embodiments of the method include, wherein:

[0045] The error function comprises the first term and a function of the scaling factor, or is given by a sum or a weighted sum of the first term and a function of the scaling factor;

[0046] The function of the scaling factor is a function which, in applications where the reference spectrum is a smooth spectrum, exhibits a clear minimum at a value of the factor corresponding to the magnitude of the interfering influences contained in the measured spectrum; and

[0047] In each application, whether the reference spectrum is a multi-peak spectrum or a smooth spectrum, the coefficient value is determined based on an error function including the first term and the function.

[0048] In certain embodiments, the method is performed only in applications where the reference spectrum is a multi-peak spectrum, or the method comprises in each application distinguishing between whether the reference spectrum is a multi-peak spectrum or a smooth spectrum. In such embodiments, in at least one or each application where the reference spectrum is a multi-peak spectrum, the error function is given by:

[0049] Item 1;

[0050] the sum or weighted sum of the first term and the second term corresponding to the error of the difference spectral function due to negative spectral intensity values ​​of the difference spectral function caused by excessive subtraction; or

[0051] The sum of the products of the first term and the penalty factor pen and the second term corresponding to the error of the difference spectral function due to negative spectral intensity values ​​of the difference spectral function caused by excessive subtraction.

[0052] In some embodiments, the method further comprises distinguishing between whether the reference spectrum is a multi-peak spectrum or a smooth spectrum in each application; and based on the distinction, determining coefficient values ​​such that when the reference spectrum is a multi-peak spectrum, the coefficient values ​​minimize the error function, and determining coefficient values ​​such that when the reference spectrum is a smooth spectrum, the coefficient values ​​minimize the alternative error function. In such an embodiment, the alternative error function:

[0053] configured to exhibit a sharp minimum at a coefficient value corresponding to the magnitude of a disturbing effect contained in the measured spectrum in an application in which the reference spectrum is a smoothed reference spectrum;

[0054] is given by the variance of the second derivative of the difference spectrum function; or

[0055] It is given by the sum of the products of a first term corresponding to an asymmetrically weighted sum of the squares of the differences between the spectral intensity values ​​of the measured spectrum and the scaled reference spectrum and a second term corresponding to the sum of the squares of the second-order derivatives of the reference spectrum at each spectral line; wherein the weighting factor applied to the squares of the differences resulting from the spectral intensity values ​​of the measured spectrum exceeding the corresponding spectral intensity values ​​of the reference spectrum is smaller than the weighting factor assigned to the squares of the differences resulting from the spectral intensity values ​​of the measured spectrum being smaller than the corresponding spectral intensity values ​​of the reference spectrum.

[0056] According to at least one embodiment, the method further comprises, in at least one or each application:

[0057] a) determine whether the reference spectrum is a multi-peak spectrum; or

[0058] b) determining whether the reference spectrum is a multimodal spectrum by performing a visual inspection of the reference spectrum, and determining that the reference spectrum is a multimodal spectrum if the reference spectrum includes at least one distinct peak; or

[0059] c) determining whether the reference spectrum is a multi-peak spectrum by: determining a baseline of the reference spectrum; determining a residual spectrum corresponding to the difference between the reference spectrum and the baseline; determining an indicator corresponding to the degree of multi-peak nature of the residual spectrum based on the residual spectrum; and performing at least one of the following: determining that the reference spectrum is a multi-peak spectrum when the indicator is greater than a predetermined threshold; and determining that the reference spectrum is a smooth spectrum when the indicator is less than a predetermined threshold.

[0060] In certain embodiments of the method according to this embodiment, the indicator is determined by determining the sum of the squares of the spectral intensity values ​​of the remaining spectrum, or by determining the ratio of the sum of the squares of the spectral intensity values ​​of the remaining spectrum divided by the average value of the spectral intensity values ​​of the remaining spectrum, or by determining the variance of the remaining spectrum.

[0061] In certain embodiments, the method further comprises determining a measurement of at least one measured variable of the medium based on the at least one or each Raman difference spectrum.

[0062] In a further embodiment, the method comprises, for at least one measured variable, determining a model for determining a measurement result of the corresponding measured variable based on a Raman difference spectrum determined by the method based on a measured spectrum of a sample of the medium exhibiting a known value of the measured variable.

[0063] The present disclosure also includes embodiments wherein:

[0064] In at least one or each application, interfering influences include at least one of: light pollution, luminescent and / or fluorescent light exhibiting at least one intensity peak; interfering analyte influences, interfering analyte composition and / or solvents contained in the medium; and background influences, and / or broadband luminescent or fluorescent light exhibiting a smooth intensity profile; and / or

[0065] Reference spectra provided include:

[0066] Using a Raman spectrometer, determining and providing at least one intensity spectrum of the interfering influence, and determining and providing a reference spectrum based on these intensity spectra; and / or

[0067] The reference spectrum is determined based on an intensity spectrum of at least one interfering analyte contained in the medium, a composition of the interfering analyte and / or a solvent.

[0068] In some embodiments, the method further comprises: using a Raman spectrometer, repeatedly determining and providing a measurement spectrum, and for at least one or each measurement spectrum, determining and providing a corresponding Raman difference spectrum; wherein:

[0069] For each measurement spectrum, determining the Raman difference spectrum includes: determining a coefficient value corresponding to the magnitude of the contribution of the interference effect contained in the corresponding measurement spectrum based on the corresponding measurement spectrum and the reference spectrum, and determining the Raman difference spectrum based on the coefficient value;

[0070] The coefficient values ​​determined based on a previously determined measurement spectrum are used to determine a Raman difference spectrum of a series of subsequently determined measurement spectra; or

[0071] The coefficient values ​​for determining the Raman difference spectrum of the continuously determined measurement spectrum are updated at least once, repeatedly or regularly by repeating the previously performed determination of the coefficient values ​​based on the newly determined measurement spectrum.

[0072] The present disclosure also includes a Raman spectrum measurement system configured to perform the method disclosed herein. The Raman spectrum measurement system includes: a Raman spectrometer configured to determine and provide a measurement spectrum of a medium; and a difference spectrum determination unit configured to determine and provide a Raman difference spectrum based on the measurement spectrum and a reference spectrum of the contribution of interference effects contained in the measurement spectrum provided to the difference spectrum determination unit.

[0073] The present disclosure also includes a non-transitory computer readable medium storing instructions that, when executed by one or more programmable processors, cause the one or more programmable processors to perform operations defining a method of Raman spectroscopy, the method comprising determining and providing a Raman difference spectrum for performing Raman spectroscopy measurements of at least one measured variable of a medium in at least one application, wherein the measured spectrum of the medium includes contributions of interfering effects that at least partially obscure spectral features associated with the measured variable(s) based on at least one measured spectrum of the medium determined by a Raman spectrometer and a reference spectrum of contributions of interfering effects included in the measured spectrum provided to the one or more programmable processors. The instructions include:

[0074] computer code for determining second derivatives of spectral lines with respect to a difference spectral function corresponding to the difference between the measured spectrum and the product of the scaling factor and the reference spectrum;

[0075] Computer code for determining a coefficient value of a scaling factor corresponding to the magnitude of the contribution of the interfering effect contained in the measured spectrum such that the coefficient value minimizes an error function representing an error of the difference spectrum function as a function of the scaling factor, wherein the error function comprises a first term corresponding to the error exhibited by the difference spectrum function due to a peak contained in the reference spectrum, the first term comprising at least one of the following:

[0076] the first sum of the areas enclosed under the second derivative of the difference spectral function in all spectral regions where the second derivative of the difference spectral function is positive;

[0077] the second sum of the areas enclosed under the second derivative of the difference spectral function in all spectral regions where the second derivative of the difference spectral function is negative; and

[0078] Computer code for determining and providing, for at least one measured spectrum determined by a Raman spectrometer, a Raman difference spectrum corresponding to the difference between the corresponding measured spectrum and the product of the coefficient value and the reference spectrum.

[0079] On the other hand, the present disclosure also includes a Raman spectroscopy measurement system, including: a Raman spectrometer, which includes: a monochromatic light source, which is configured to emit excitation light with a predetermined excitation wavelength to a measurement area of ​​a sample configured to accommodate a medium to irradiate the sample; and a spectroscopic measurement unit, which is configured to receive Raman scattered light emitted from the irradiated sample and provide a measurement spectrum, including an intensity spectrum of the Raman scattered light, wherein the spectroscopic measurement unit includes a disperser, a detector and a signal processor, and the signal processor is configured to determine and provide a spectral intensity value of the measurement spectrum based on a signal from the detector; and a difference spectrum determination unit, which is configured to determine a Raman difference spectrum based on the measurement spectrum, wherein the signal processor of the spectroscopic measurement unit and the difference spectrum determination unit are cooperatively configured to perform the method according to the present disclosure.

[0080] In at least one embodiment, such a Raman spectroscopy measurement system further comprises a signal processing unit configured to determine and provide a measurement result based on the Raman difference spectrum provided by the difference spectrum determination unit and based on a previously determined model for determining a measurement result of at least one measured variable of the medium based on spectral intensity values ​​of the Raman difference spectrum. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] The described embodiments and other features, advantages and disclosures contained herein and the manner in which they are achieved will become apparent and the present disclosure will be better understood by referring to the following description of various embodiments of the present disclosure in conjunction with the accompanying drawings, in which:

[0082] Figure 1 A flow chart showing a method of Raman spectroscopy according to the present disclosure is shown;

[0083] Figure 2 A Raman spectroscopy measurement system according to the present disclosure is shown;

[0084] Figure 3 An exemplary intensity profile of a measured spectrum is shown;

[0085] Figure 4 An exemplary intensity profile of a reference spectrum exhibiting a constant baseline is shown;

[0086] Figure 5 An exemplary intensity profile of a reference spectrum exhibiting a smooth baseline is shown;

[0087] Figure 6 The difference spectral function determined by degree subtraction is shown;

[0088] Figure 7 Shows Figure 6 The second derivative of the difference spectrum function shown;

[0089] Figure 8Shows that by applying Figure 3 The difference spectrum function is determined by the coefficient value of the scaling factor corresponding to the amplitude of the interference effect contained in the measured spectrum shown;

[0090] Fig. 9 Shows Figure 8 The second derivative of the difference spectrum function shown;

[0091] Fig.10 The difference spectral function determined by oversubtraction is shown;

[0092] Fig.11 Shows Fig.10 The second derivative of the difference spectrum function shown; and

[0093] Fig.12 The error function is shown. DETAILED DESCRIPTION

[0094] The present disclosure includes a method of Raman spectroscopy, comprising determining and providing a Raman difference spectrum for performing Raman spectroscopic measurements of at least one measured variable of a medium in at least one application, wherein the measured spectrum of the medium Contributions of interfering effects are included, for example, contributions that at least partially obscure spectral features associated with the measured variable(s). Figure 1 A flow chart of a method 200 according to the present disclosure employed in each application is shown.

[0095] Figure 2 The Raman spectroscopy measurement system 100 for performing the method 200 is shown. The Raman spectroscopy measurement system 100 comprises a Raman spectrometer 110 configured to determine and provide a measurement spectrum of a medium. and a difference spectrum determination unit 120, which is configured to be based on the measured spectrum To determine and provide the corresponding Raman difference spectrum , which are configured together to execute Figure 1 Method 200 is shown.

[0096] Figure 2 The exemplary Raman spectrometer 110 shown includes a monochromatic light source 1, such as a laser, which is configured to emit a light having a predetermined excitation wavelength The excitation light L0 is emitted to a measurement region 3 configured to accommodate a sample SM of a medium.

[0097] In some embodiments, for example, the excitation wavelength is a wavelength in the visible light or near infrared wavelength range, such as 785 nm or 532 nm. In some embodiments, the Raman spectrometer 110 includes a filter 5 , such as a notch filter, configured to filter out the Raman scattered light LR included in the light L1 emitted from the measurement region 3 .

[0098] The Raman spectrometer 110 further comprises a spectrometry unit 7, which is configured to receive the Raman scattered light LR emitted from the illuminated sample SM. The spectrometry unit 7 is configured to determine and provide a measurement spectrum . Measure the spectrum For example, the Raman scattered light LR is provided in a predetermined spectral range. The intensity spectrum within, for example, consists of The wavelength range or wave number range of the wave number shift of the Raman scattered light caused by the excitation light L0.

[0099] In some embodiments, the spectrometry unit 7 includes, for example: a scatterer 9 (e.g., a diffraction grating or a holographic grating) that scatters the incident measurement light LR; a detector 11 that receives the scattered Raman scattered light LR; and a signal processor 13 (e.g., a microprocessor) that is connected to the detector 11. The detector 11 is configured to determine and provide a detector signal corresponding to the spectral intensity of the incident scattered measurement light LR. The signal processor 13 is configured to determine and provide a measurement spectrum based on the detector signal. The spectral intensity value .

[0100] The difference spectrum determination unit 120 is, for example, a computer, a microprocessor or other type of programmable computing device, which includes a memory (e.g., a non-transitory computer-readable medium), and is configured to determine the difference spectrum based on the measured spectrum by executing a computer program (e.g., processor-executable instructions) installed on the difference spectrum determination unit 120. To determine the Raman difference spectrum .

[0101] In each application, the method 200 comprises a method step A1 of determining a measurement spectrum of the medium using the Raman spectrometer 110. , and method step A2, which provides for measuring the spectrum The reference spectrum contains the contributions of the interfering effects .

[0102] Depending on the application, the measured spectrum Interfering influences that come into play include, for example, light pollution, for example, luminescent light and / or fluorescent light that exhibits at least one intensity peak, and / or interfering influences of the composition of the analyte and / or the solvent contained in the medium that interferes with the analyte. In addition or as an alternative, interfering influences include, for example, background influences, for example, broadband luminescence or fluorescent light that exhibits at least approximately a smooth intensity profile.

[0103] As used in this disclosure, the term "peak" refers to a local maximum of a spectrum or function, which may have multiple local maxima, and the term "valley" refers to a local minimum of a spectrum or function, which may have multiple local minima. Thus, "multimodality" refers to the relative prevalence of local maxima in a spectrum or function.

[0104] In certain embodiments, a reference spectrum is provided For example, it includes using a Raman spectrometer to determine and provide at least one intensity spectrum of the interference effect, and determining and providing a reference spectrum based on these intensity spectra. .

[0105] As an example, in some embodiments, a reference spectrum is determined For example, it comprises determining at least one intensity spectrum of (multiple) interfering analytes and / or solvents contained in the medium. The determination of each intensity spectrum is performed, for example, by a high-precision spectroscopic measuring instrument and / or under laboratory conditions.

[0106] Reference spectrum For example, it is determined and / or provided as a normalized spectrum, which represents a reference spectrum The spectral intensity profile on a normalized intensity scale, e.g., by The spectral intensity values ​​are determined relative to the reference spectrum The relative intensity value of the maximum intensity value contained in .

[0107] Figure 3 The measured spectrum determined in method step A1 is shown An exemplary intensity profile of . Figure 4 Shows Figure 3 The measured spectrum shown The reference spectrum contains the contributions of the interfering effects An exemplary intensity profile of .

[0108] In the example shown, the reference spectrum It is a multi-peak spectrum, which shows multiple different peaks Pr1, Pr2, Pr3 superimposed on a constant baseline Although the method 200 disclosed herein is subsequently based on Figure 3 and Figure 4An exemplary intensity profile is shown , , but the method 200 can be used in which the reference spectrum Shows a smooth baseline (e.g., a baseline caused by background effects (e.g., broadband luminescent or fluorescent light that exhibits a smooth intensity profile—e.g., a profile that includes gradually curved segments) ) in the same way. Figure 5 shows the performance superimposed on a smooth baseline The reference spectrum of the different peaks P The intensity profile .

[0109] As described in the introduction to this disclosure, by measuring the spectrum Subtract the correctly scaled reference spectrum Determine and provide Raman difference spectrum (For example, through ,in, is the coefficient value of the scaling factor s), which can significantly reduce or even eliminate the negative effects of interference on Raman spectroscopy measurement.

[0110] Eliminate the corresponding measured spectrum by spectral subtraction The contribution of the interference effect included in the scaling factor s is required to determine the coefficient value , so that it is consistent with the corresponding measured spectrum The magnitude of the contribution of the interference effect contained in corresponds to the value of the first coefficient of the scaling factor s which is too small. Insufficient subtraction and by adopting a second coefficient value that is too large for the scaling factor s Subtract the Raman difference spectrum obtained by modification , but none of them removes it.

[0111] The disclosed method recognizes the contribution of interference effects to the measured spectrum The effect depends on the reference spectrum The spectral shape and the coefficient of determination value Due consideration must be given to the possible inclusion of The effect of peaks Pr1, Pr2, and Pr3.

[0112] In this respect, the method 200 comprises a method step A3 for determining a difference spectral function G corresponding to the measured spectrum With the scaling factor s and the reference spectrum The difference between the products of .

[0113] In some embodiments, method 200 further includes filtering the measured spectrum and reference spectrum In such an embodiment, based on the filtered measured spectrum and the filtered reference spectrum To determine the difference spectrum function G, for example, by Subtract the scaling factor s from the filtered reference spectrum is determined by multiplying to be sure.

[0114] In method step F, the measured spectrum is filtered, for example, using a smoothing filter and reference spectrum To perform filtering. In some embodiments, the smoothing filter is, for example, a second-order or higher-order smoothing filter, or a Savitzky-Golay filter, for example, a second-order or higher-order Savitzky-Golay filter, for example, a Savitzky-Golay filter with a window width of 20 to 30 (e.g., a window width of 25) and a polynomial order of 3 or higher. Alternatively, another type of smoothing filter, such as a wavelet decomposition filter, may be used instead.

[0115] The advantage of filtering is that it eliminates and reference spectrum The noise contained in , which might otherwise corrupt the coefficient values determination, for example, due to the error caused by the noise of the difference spectrum function G.

[0116] At the same time, filtering also offers the advantage of preserving the spectral shape of the corresponding spectrum. The spectral intensity profile of The spectral intensity profile of is at least approximately the same as the reference spectrum after filtering The spectral intensity profile of The spectral intensity profile of At least approximately the same.

[0117] In the method 200 disclosed herein, determining the corresponding measured spectrum The coefficient value of the scaling factor s corresponding to the magnitude of the contribution of the interference effect contained in The method comprises a step A4 of determining the difference spectral function G with respect to the spectral line The second derivative SDG of the measured spectrum The coefficient value of the scaling factor corresponding to the magnitude of the contribution of the interference effect contained in , so that the coefficient value The error function E(s) representing the error of the difference spectral function G as a function of the scaling factor s is minimized.

[0118] In method step A4, the second derivative SDG of the difference spectrum function G is obtained, for example, by The second derivative of Subtract the scaling factor s from the filtered reference spectrum The second derivative of is determined by multiplying to be sure.

[0119] The Savitsky-Golay filter is used to measure the spectrum. and reference spectrum In the embodiment in which filtering is performed, the filtering performed in method step F and the determination of the second-order derivative SDG performed in method step A4 are performed, for example, in a combined single method step, which performs a second-order Savitsky-Golay differentiation on the difference spectral function G, which is obtained by dividing the difference spectral function G by the unfiltered measured spectrum. Subtract the scaling factor s from the unfiltered reference spectrum is determined by multiplying to be sure.

[0120] In method step A5, the reference spectrum The effect of the peaks Pr1, Pr2, Pr3 contained in the reference spectrum is explained by the error function E(s), which includes a first term T1 corresponding to the peaks Pr1, Pr2, Pr3 contained in the reference spectrum. The error in the difference spectrum function G caused by the included peaks Pr1, Pr2, and Pr3.

[0121] Then based on Figure 6-11 Describing the nature of this error and its dependence on the value of the scaling factor s, the Figure 6-11 An exemplary difference spectrum function is shown , , and its second-order derivative , , The main shapes of these shapes are based on Figure 3 The measured spectrum shown and Figure 4 Reference spectrum shown And sure.

[0122] exist Figure 3 The measured spectrum shown The intensity profile In the measurement spectrum The amplitude and spectral shape of the contribution of the interference effects contained in are indicated by the shaded areas. Each shaded area corresponds to the reference spectrum One of the peaks Pr1, Pr2, Pr3. Figure 4 Reference spectrum shown The exemplary peaks Pr1, Pr2, Pr3 relative to the measured spectrum Other spectral features Pm1, Pm2 (eg, peaks) included in appear at different positions.

[0123] like Figure 3 As shown, the reference spectrum The first peak Pr1 appears in the first spectral region In which the measured spectrum This first peak Pr1 makes the measured spectrum In the first spectral region The reference spectrum includes an additional first maximum value Max(Pr1). The second peak Pr2 appears in the second spectral region In the second spectral region With spectral region Partial overlap, where the measured spectra Including the first spectral feature Pm1. Correspondingly, the second peak Pr2 makes the measured spectrum The reference spectrum includes an additional second maximum value Max(Pr2) superimposed on one side of the first spectral feature Pm1. The third peak Pr3 appears in the covered spectral region The third spectral region In which the measured spectrum In the example shown, the third peak Pr3 makes the measured spectrum The third maximum value Max(Pr3) exhibits a maximum intensity exceeding the peak intensity of the second spectral feature Pm2 and exceeding the third spectral region The peak width of the peak width of the second spectral feature Pm2 in.

[0124] Figure 6 shows that the measured spectrum is less than The coefficient value corresponding to the amplitude of the interference effect contained in The first coefficient value of Difference spectral function at . Figure 7 The first coefficient value is shown in Difference spectral function at The second derivative of . Too small first coefficient value It will lead to insufficient subtraction.

[0125] Figure 8 shows the measured spectrum The coefficient value corresponding to the amplitude of the interference effect contained in Difference spectral function at . Fig. 9 Shows the coefficient value at Difference spectral function at The second derivative of .

[0126] Fig.10 shows that the measured spectrum is greater than The coefficient value corresponding to the amplitude of the interference effect contained in The second coefficient value of Difference spectral function at . Fig.11 The second coefficient value is shown in Difference spectral function at The second derivative of . Too large a second coefficient value Will lead to excessive subtraction.

[0127] like Figure 6 and Figure 7 As shown, the reference spectrum The insufficient subtraction of the first peak Pr1 reduces the significance of the first maximum value Max(Pr1), but fails to eliminate the first maximum value Max(Pr1). Figure 7 The second derivative shown The valley Vsd is shown twice, which corresponds to Figure 6 The difference spectrum function shown In the first spectral region The inflection point in is surrounded by two peaks Psd.

[0128] like Figure 8 and Fig. 9 As shown, application and measurement spectrum The magnitude of the interference effect contained in the scaling factor s corresponds to the value of the difference spectrum function that eliminates Reference spectrum in Correspondingly, the second-order derivative In the first spectral region It does not show any peaks or valleys.

[0129] Oversubtraction of reference spectra The first peak Pr1 makes Fig.10The difference spectrum function shown shows a minimum value Min(Pr1). This makes Fig.11 The second derivative shown The peak Psd is shown twice, which corresponds to Fig.10 The difference spectrum function shown The first spectral region of the inflection point The two valleys in the middle are surrounded by Vsd.

[0130] In a similar manner, the reference spectrum The partially overlapping second peak Pr2 is insufficiently subtracted to reduce Figure 6 The significance of the second maximum value Max(Pr2) shown in the figure is too great, and excessive subtraction of the second peak Pr2 will Fig.10 The difference spectrum function shown It shows the minimum value Min(Pr2). Figure 7 , 9 The second-order derivatives shown in and 11 are , , It can be clearly seen from the comparison of the reference spectrum The under-subtraction and over-subtraction of the second peak Pr2 make the second derivative , exhibits additional peaks Psd and valleys Vsd when applied to the measured spectrum The coefficient value corresponding to the amplitude of the interference effect contained in When these peaks Psd and valleys Vsd are eliminated, such as Fig. 9 shown.

[0131] Relative to the reference spectrum The completely overlapping third peak Pr3, such as Figure 6 As shown, the third peak Pr3 is subtracted and Fig.10 As shown in the figure, excessive subtraction of the third peak Pr3 will affect the third spectral range. The corresponding difference spectrum function in , The multi-peak spectral shape of Figure 7 and 11 The corresponding second-order derivative is shown , exhibits additional peaks Psd and valleys Vsd when applied to the measured spectrum The coefficient value corresponding to the amplitude of the interference effect contained in When these peaks Psd and valleys Vsd are eliminated, such as Fig. 9 shown.

[0132] As is apparent from this example, for a multi-peak reference spectrum Over-subtraction and under-subtraction will cause the second-order derivative SDG to show additional peaks Psd and valleys Vsd. The peaks Pr1, Pr2, and Pr3 contained in the measured spectrum These additional peaks Psd and valleys Vsd will appear regardless of the positions of the other spectral features Pm1 and Pm2 contained in .

[0133] Take the first coefficient value As a starting point, gradually increase the value of the scaling factor s until the correct coefficient value , reducing the significance of the extra peak Psd contained in the second-order derivative SDG caused by the incorrect size of the scaling factor s, so that when the scaling factor s reaches the correct coefficient value These peaks Psd are eliminated. Further increasing the scaling factor s beyond the correct factor value The value of , causes the extra peak Psd contained in the second-order derivative SDG due to insufficient subtraction to be transformed into the valley Vsd exhibited by the second-order derivative SDG due to excessive subtraction. Similarly, with the first coefficient value As a starting point, gradually increase the value of the scaling factor s until the correct coefficient value , reducing the significance of the extra valley Vsd contained in the second-order derivative SDG due to the incorrect size of the scaling factor s, so that when the scaling factor s reaches the correct coefficient value Further increasing the scaling factor s beyond the correct factor value The value of , causes the extra valley Vsd contained in the second-order derivative SDG due to insufficient subtraction to be transformed into the peak Psd exhibited by the second-order derivative SDG due to excessive subtraction.

[0134] This effect makes the reference spectrum The error of the difference spectral function G caused by the peaks Pr1, Pr2, and Pr3 contained therein can be quantitatively determined as a function of the scaling factor s based on: in the spectral region where the second-order derivative SDG exhibits a peak Psd, based on the size of the area enclosed below the second-order derivative SDG; in the spectral region where the second-order derivative SDG exhibits a valley Vsd, based on the size of the area enclosed below the second-order derivative SDG.

[0135] Correspondingly, the first term T1 of the error function E(s) comprises a first sum S1 given by the sum of the areas enclosed below the second-order derivative SDG of the difference spectral function G in all spectral regions in which the second-order derivative SDG exhibits a peak Psd. These spectral regions are given by spectral regions in which the second-order derivative SDG of the difference spectral function G is positive.

[0136] In addition or as an alternative, the first term T1 of the error function E(s) comprises a second sum S2 given by the sum of the areas enclosed below the second derivative SDG of the difference spectral function G in all spectral regions in which the second derivative SDG exhibits a valley Vsd. These spectral regions are given by spectral regions in which the second derivative SDG of the difference spectral function G is negative.

[0137] In some embodiments, the first sum S1 is determined, for example, as an integral, for example, of the spectral value of the second-order derivative SDG The corrected linear unit relU in the spectral range The trapezoidal integral on trapz, for example, is obtained by , where, for all spectral values ​​with positive values ​​greater than or equal to zero , spectral value The corrected linear unit relU (e.g., ramp function) is equal to the spectral value The absolute value of , and for all spectral values ​​with negative values , spectral value The rectified linear unit relU is equal to zero.

[0138] Similarly, the second sum S2 is determined, for example, as the integral, for example, of the spectral value of the second derivative SDG The negative one times the rectified linear unit in the spectral range The trapezoidal integral on trapz, for example, is obtained by .

[0139] In some embodiments, the first term T1 is determined, for example, as a term T1a, which is given by the sum of the first sum S1 and the second sum S2. In this case, the first term T1 is determined, for example, as an integral, for example, of the spectral value of the second derivative SDG The absolute value abs( ) in the spectral range The trapezoidal integral on trapz, for example, is obtained by .

[0140] As an alternative, in some embodiments, the first term T1 is determined, for example, as a term given by a weighted sum of the first sum S1 and the second sum S2.

[0141] As an example, in these embodiments, the first term T1 is determined as term T1b, which is the sum of the first and S1 and the penalty factor The sum of the products of the second and S2 is given, for example, by .

[0142] As another example, the first term T1 is determined, for example, as a term T1c given by the sum of two products, wherein the first sum S1 is multiplied by the penalty factor The first product is determined by multiplying 1 with the penalty factor The difference between the second and S2 is multiplied to determine the second product, for example, by .

[0143] In which the reference spectrum For multi-peak spectroscopy applications, since the reference spectrum The spectral values ​​of the additional peaks Psd and valleys Vsd of the second-order derivative SDG caused by over-subtraction or under-subtraction of the peaks Pr1, Pr2, and Pr3 included in , with scaling factor s and corresponds to the measured spectrum The coefficient value of the magnitude of the contribution of the interference effect contained in This results in the first term T1 being a V-shaped function of the scaling factor s, which corresponds to the measured spectrum The coefficient value of the magnitude of the contribution of the interference effect contained in There is a clear minimum The V-shape provides the advantage of being able to determine exactly the value of the coefficient that minimizes the first term T1 .

[0144] Thus, in some embodiments, the error function E(s) is given, for example, by the first term T1, for example, by This is given. Fig.12 It is shown in Fig.12 Shown based on Figure 3 The measured spectrum shown and Figure 4 Reference spectrum shown An exemplary embodiment of a determined error function E(s), wherein the error function E(s) is given by a first term T1.

[0145] Alternatively, in some embodiments, the error function E(s) used in method step A5 comprises, for example, at least one additional term. In these embodiments, the error function E(s) is, for example, given by the sum or weighted sum of the first term T1 and each additional term.

[0146] As an example, in some embodiments, the error function E(s) comprises, for example, a second term T2 corresponding to negative spectral intensity values ​​of the difference spectral function G due to over-subtraction. The resulting error in the difference spectrum function G.

[0147] Fig.10 Such an example is shown in FIG. 1 , where the oversubtraction of each of the three peaks Pr1 , Pr2 , Pr3 results in a spectral intensity value of the difference spectral function G In which the scaling factor With reference spectrum The spectral intensity value of the product is greater than the measured spectrum The corresponding spectral intensity value is negative in the spectral region. Fig.10 These regions are where the difference spectrum function G exhibits minimum values ​​Min(Pr1), Min(Pr2), , spectral region.

[0148] The result is a difference spectrum function G in which the scaling factor s is the same as the reference spectrum The spectral intensity value of the product is greater than the measured spectrum The corresponding spectral intensity value The spectral region shows negative spectral intensity values The same effect of over-subtraction also occurs when the reference spectrum For applications where spectrum smoothing is desired.

[0149] Regardless of the reference spectrum Whether it is multi-peaked or smooth, the second term T2 is determined, for example, as the spectral intensity value of the difference spectral function G in which The sum of the areas enclosed under the difference spectrum function G in all spectral regions where the difference spectrum function G is negative. In some embodiments, the second term T2 is determined as an integral, for example, the spectral intensity value of the difference spectrum function G The trapezoidal integral of the negative-one rectified linear unit over the spectral range is trapz, for example, by Sure.

[0150] In some embodiments, the error function E(s) is given, for example, by the sum of the product of the first term T1 and the penalty factor pen and the second term T2, for example, by Given.

[0151] Including the second term T2 is particularly advantageous in applications in which even in principle negligible oversubtractions are to be avoided, for example because of the coefficient values Determined Raman difference spectrum Further data processing cannot cope with Raman difference spectroscopy Negative spectral intensity values ​​are ignored, even if they are negligibly small.

[0152] Considering the second derivative SDG of the difference spectrum function G, where the difference spectrum function G is a constant or a spectral line is zero in all spectral regions of the linear function of , and the second term T2 itself cannot prevent insufficient subtraction, embodiments in which the error function E(s) is given by the first term T1 and embodiments in which the error function E(s) is given by the sum or weighted sum of the first term T1 and the second term T2 (e.g., ) is preferably used only where the reference spectrum For multi-peak spectroscopy applications.

[0153] This can be ensured in various ways. In certain embodiments, this can be achieved, for example, by limiting the application of the method 200 disclosed herein to applications where the reference spectrum This is achieved through the application of multi-peak spectroscopy.

[0154] In other embodiments, the method 200 includes, for example, a method step K of distinguishing the reference spectrum in each application Is it a multi-peak spectrum or a smooth spectrum.

[0155] Based on this distinction, in the reference spectrum In the case of a multi-peak spectrum, the coefficient value of the scaling factor s is For example, it is determined based only on the error function E(s) given by the first term T1 or based on the error function E(s) given by the sum or weighted sum of the first term T1 and the second term T2.

[0156] As an option, such an embodiment of the method 200 comprises, for example, a further method step A5', which comprises: In the case of a smoothed spectrum, the coefficient of determination is , so that it minimizes the substitution error function RE(s).

[0157] For example, the replacement error function RE(s) is configured such that it refers to the spectrum In applications for smoothing spectra, the coefficient value corresponding to the amplitude of the interference effects contained in the measured spectrum There is a clear minimum value.

[0158] With respect to the substitute error function RE(s), the function described in the introduction of the present disclosure may be employed. As an example, a substitute error function RE(s) given by the variance of the second derivative SDG of the difference spectrum function G may be employed. Another example is a substitute error function RE(s) given by the sum of the product of the first term and the penalty factor q and the second term, the first term corresponding to the measured spectrum The spectral intensity value With scaled reference spectrum The second term corresponds to the asymmetrically weighted sum of the squares of the differences between Reference spectrum at The second derivative of The sum of the squares of , for example, is given by:

[0159]

[0160] Among them, applied to the measurement spectrum The spectral intensity value Exceeding the reference spectrum The corresponding spectral intensity value The weighting factor of the square of the difference is significantly smaller than that assigned to the measured spectrum The spectral intensity value Smaller than the reference spectrum The corresponding spectral intensity value The weighting factor of the square of the difference is .

[0161] Relative to the reference spectrum Is it multi-peaked or smooth? Step K: For some applications, the reference spectrum can be known in advance. Whether it is a multi-peak spectrum. In this case, for example, differentiation is performed based on this prior knowledge.

[0162] As an alternative, in certain embodiments, method step K comprises, for example, determining a reference spectrum Whether it is a multi-peak spectrum.

[0163] In such an embodiment, method step K comprises, for example, performing a reference spectrum The intensity profile and, if it includes at least one distinct peak Pr1, Pr2, Pr3, determine the reference spectrum It is a multi-peak spectrum.

[0164] As an alternative, in certain embodiments, method step K comprises, for example, determining a reference spectrum The baseline ( ), and determine the reference spectrum and baseline ( ) corresponds to the residual spectrum R, for example, by After this determination, an index KP corresponding to the degree of multimodality of the residual spectrum R is determined.

[0165] In some embodiments, for example, based on the spectral intensity value of the residual spectrum R The square sum of KP is used to determine the indicator KP, for example, as the spectral intensity value The sum of the squares divided by the spectral intensity value The average value of the RA ratio, for example, is determined by the following formula:

[0166]

[0167] As an alternative, in certain embodiments the indicator KP is determined, for example, either based on the variance of the residual spectrum or based on another function of the residual spectrum R corresponding to the degree of its multimodality.

[0168] After determining the index KP, if the index KP is greater than the predetermined threshold Kr, the reference spectrum is determined. For a multi-peak spectrum and / or when the indicator KP is less than a predetermined threshold Kr, determine the reference spectrum To smooth the spectrum.

[0169] However, distinguishing the reference spectrum multimodal or smooth requires at least some additional effort, for example due to the need for additional computing power and / or increased complexity of the computer program executed by the difference spectrum determination unit 120, which is based on the measured spectrum provided to the difference spectrum determination unit 120. and reference spectrum To perform method 200.

[0170] This burden can be avoided by defining the error function E(s) used in method step A5 so that it can be used both for the reference spectrum and for For multi-peak spectroscopy applications, it can also be used for reference spectra For smoothing spectrum applications.

[0171] In some embodiments, this is achieved, for example, by an error function E(s) comprising a first term T1 and an additional function f(s) of the scaling factor s, wherein the reference spectrum For smoothing spectrum applications in the measurement spectrum The coefficient value corresponding to the amplitude of the interference effect contained in In such an embodiment, the error function E(s) is determined, for example, as the sum or weighted sum of the first term and the additional function f(s).

[0172] In some embodiments, the error function E(s) including the first term T1 and the function f(s) is given by, for example, a function including the first term T1 and the second term T2 and an additional third term T3 corresponding to the error of the difference spectrum function G due to insufficient subtraction. The third term T3 is, for example, a function including the spectral intensity value of the difference spectrum function G The sum of the areas enclosed under the difference spectral function G in all spectral regions where is positive gives. In this regard, the third term T3 is determined, for example, as the spectral intensity value of the difference spectral function G The rectified linear unit in the spectral range The integral on (e.g., trapezoidal integral trapz), for example, by .

[0173] In such an embodiment, the error function E(s) is determined, for example, as the sum or weighted sum of the first term T1, the second term T2 and the third term T3, for example by ,in, is the penalty factor applied to the second term T2 to penalize excessive subtractions.

[0174] The error function E(s) includes the first term T1 and the function f(s). For example, the function f(s) is given by Given, it offers the advantage that, regardless of the reference spectrum Whether it is a multi-peak spectrum or a smoothed spectrum, it can be used in method step A5 in each application. This offers the advantage that in each application, it is neither necessary to know nor to determine the reference spectrum. Whether it is a multi-peak spectrum.

[0175] Using the error function E(s) comprising the first term T1 and the function f(s) in each application provides a simplified installation in the difference spectrum determination unit 120 (eg Figure 2 The execution coefficient value on The function f(s) comprising the second term T2 and the third term T3 offers the advantage that only very little computing power is required to determine the two additional terms T2, T3. The error function E(s) comprising the function f(s) also offers the advantage that the reference spectrum can be covered without replacing the error function RE(s). Advantages of non-multimodal applications and no need for multimodal and smooth reference spectra Distinguishing between Figure 1 The method 200 shown in FIG. 1 does not require the determination of a reference spectrum. Method step K for determining whether it is a multi-peak spectrum.

[0176] Regardless of whether the error function E(s) used in method step A5 comprises only the first term T1, the first term T1 and the second term T2, or the first term T1 and the function f(s), the coefficient value of the scaling factor s are determined so that they are equal to the error function E(s) exhibiting its minimum value The value of the scaling factor s at this time is Min[E(s)].

[0177] In some embodiments, in method step A5, the coefficient value of the error function E(s) is minimized. For example, it is determined by performing a golden section search, a particle swarm optimization method, or a gradient descent method. Alternatively, another method for determining the minimum value of a function may be used.

[0178] The coefficient of determination Afterwards, the method 200 further comprises a method step A6, which is to determine at least one measurement spectrum determined by the Raman spectrometer 110 , determine and provide Raman difference spectra , the Raman difference spectrum Corresponding to the measured spectrum With coefficient value and reference spectrum The difference between the products of .

[0179] In some embodiments, the method 200 includes, for example, repeatedly determining and providing a measurement spectrum using the Raman spectrometer 110. , and for each measured spectrum , determine and provide the corresponding Raman difference spectrum .

[0180] In this case, for each measured spectrum , for example based on the corresponding measured spectrum , by determining the coefficient values ​​as described above To determine the Raman difference spectrum , and therefore based on the determined coefficient values , determine the Raman difference spectrum This measures the spectrum This is particularly advantageous in applications where the magnitude of the contribution of the contained interfering effects may vary rapidly.

[0181] Alternatively, in some embodiments, based on a previously determined measured spectrum Determined coefficient value For example, to determine a series of subsequently determined measured spectra Raman difference spectroscopy In such an embodiment, the coefficient value For example, the measured spectrum is used for all subsequent determinations As an alternative, based on the newly determined measured spectrum , by repeating the previously performed determination coefficient value Methods such as updating the coefficient values ​​at least once, repeatedly or periodically The latter measures the spectrum The magnitude of the contribution of the interference effects contained in the measured spectrum may be determined in a continuous manner. This is particularly advantageous in applications where changes occur over long time scales compared to the time intervals between them.

[0182] Whether determining and / or updating coefficient values What is the frequency of the Raman difference spectrum determined and provided in method step A6? Both can be used in various ways.

[0183] As an example, in some embodiments, the method 200 disclosed herein includes, for example, a method step A7, which is based on at least one or each Raman difference spectrum determined in method step A6. , a measurement result MR of at least one measured variable of the medium is determined. In this respect, Figure 2 The Raman spectrum measurement system 100 shown in the figure includes, for example, a signal processing unit 130, such as a computer, a microprocessor or other types of computing units, based on the Raman difference spectrum provided by the difference spectrum determination unit 120. As well as based on the previously determined Raman difference spectroscopy The spectral intensity values ​​of the measurement result MR are determined by a model MOD of the measurement result MR to determine and provide the measurement result MR.

[0184] Additionally or alternatively, in certain embodiments, method 200 comprises, for example, method step A8, which for at least one measured variable is based on a measured spectrum of a sample SM of a medium exhibiting a known value of the corresponding measured variable as described above. The Raman difference spectrum , determine a model MOD for determining the measurement result MR of the corresponding measured variable. Determining the model MOD includes, for example, based on Raman difference spectroscopy and corresponding known values ​​of the corresponding measured variables, identifying spectral features associated with the measured variables and / or relationships between spectral features associated with the measured variables, and based on Raman difference spectroscopy The selected spectral intensity values ​​of are used to determine and provide an algorithm for calculating a measurement result MR of the measured variable. In certain embodiments, at least one model MOD determined in this way is then used, for example, in method step A7 to determine measurement result(s) MR.

[0185] Based on Raman difference spectroscopy Determining a measurement result MR and / or determining at least one model MOD for determining a measurement result MR provides the advantage that in Raman difference spectroscopy The spectral information of the measured variable contained in the measured spectrum is no longer Thus, an accurate model MOD and / or a high measurement accuracy of the measurement result MR is achieved.

Claims

1. A method (200) of Raman spectroscopy for performing Raman spectroscopic measurements of at least one measured variable of a medium in at least one application, wherein: The measured spectrum of the medium ( ) comprises at least partially obscuring the contribution of interfering effects of spectral features associated with the at least one measured variable, the method comprising, in at least one application: Determine at least one measured spectrum ( ); Provide the corresponding measured spectrum ( ) contains the contribution of the interference effect to the reference spectrum ( ); Determine the corresponding measured spectrum ( ) and scaling factor (s) with the reference spectrum ( ) corresponds to the difference spectral function (G); Determine the difference spectral function (G) relative to the spectral line ( )’s second-order derivative (SDG); Determine the corresponding measured spectrum ( ) corresponds to the coefficient value ( ), so that the coefficient value ( ) minimizes an error function (E(s)) representing the error of the difference spectrum function (G) as a function of the scaling factor (s), wherein the error function (E(s)) comprises a first term (T1) corresponding to the difference spectrum function (G) due to the reference spectrum ( ) contained in the peaks (Pr1, Pr2, Pr3) and the error exhibited by the difference spectrum function (G), the first term (T1) includes at least one of the following: a first sum (S1) of the areas enclosed below the second-order derivative (SDG) of the difference spectral function (G) in all spectral regions where the second-order derivative (SDG) of the difference spectral function (G) is positive; and a second sum (S2) of the areas enclosed below the second derivative (SDG) of the difference spectral function (G) in all spectral regions where the second derivative (SDG) of the difference spectral function (G) is negative; and Determine and provide the corresponding measured spectrum ( ) and the coefficient value ( ) and the reference spectrum ( ) corresponds to the Raman difference spectrum ( ).

2. The method (200) according to claim 1, further comprising: The corresponding measured spectrum ( ) and the reference spectrum ( ) for filtering; and Based on the filtered measured spectrum ( ) and the filtered reference spectrum ( ) to determine the difference spectral function (G) and / or the second derivative (SDG) of the difference spectral function (G).

3. The method (200) of claim 1, wherein: The first term (T1) is determined as the sum or weighted sum of the first sum (S1) and the second sum (S2), the integral or trapezoidal integral of the absolute value of the spectral value of the second derivative (SDG) of the difference spectral function (G) over the spectral range, the integral of the first sum (S1) and the penalty factor ( ) and the second sum (S2) gives a term (T1b), or a term (T1c) given by the sum of two products, where the first sum (S1) is combined with a penalty factor ( ) to determine the first product, and by adding one to the penalty factor ( ) is multiplied by the second sum (S2) to determine a second product; The first sum (S1) is determined as the spectral value ( ) of the rectified linear unit in the spectral range ( ) or trapezoidal integrals over ; and / or The second sum (S2) is determined as the spectral value of the second derivative (SDG) ( ) times the negative of the rectified linear unit in the spectral range ( ) or trapezoidal integral.

4. The method (200) according to claim 1, wherein: The error function (E(s)) also includes a second term (T2), and the second term (T2): corresponds to the negative spectral intensity values ​​( ) caused by the error of the difference spectrum function (G); The spectral intensity value ( ) is given by the sum of the areas enclosed beneath the difference spectral function (G) in all spectral regions where ) is negative; and / or The spectral intensity value ( ) times the negative of the rectified linear unit in the spectral range ( ) or trapezoidal integral.

5. The method (200) according to claim 4, wherein: The error function also includes a third term (T3), and the third term (T3): corresponds to an error in said difference spectral function (G) due to insufficient subtraction; The spectral intensity value ( ) is given by the sum of the areas enclosed beneath the difference spectral function (G) in all spectral regions where ) is positive; and / or The spectral intensity value ( ) of the rectified linear unit in the spectral range ( ) or trapezoidal integral.

6. The method (200) according to claim 5, wherein: The error function (E(s)) is given by: the sum or weighted sum of the first term (T1), the second term (T2) and the third term (T3); or The first term (T1), the penalty factor ( ) multiplied by the second term (T2), and the sum of the third term (T3).

7. The method (200) according to claim 6, wherein: In each application, regardless of the reference spectrum ( ) is a multi-peak spectrum or a smooth spectrum, the coefficient value ( ).

8. The method (200) according to any one of claims 1 to 3, wherein: The error function (E(s)) comprises the first term (T1) and a function (f(s)) of the scaling factor (s), or is given by a sum or a weighted sum of the first term (T1) and a function (f(s)) of the scaling factor (s); The function (f(s)) of the scaling factor (s) is where the reference spectrum ( ) is applied for the smoothed spectrum in comparison with the corresponding measured spectrum ( ) corresponds to the coefficient value ( ) shows a clear minimum at as well as In each application, regardless of the reference spectrum ( ) is a multi-peak spectrum or a smooth spectrum, the coefficient value ( ).

9. The method (200) according to any one of claims 1 to 3, wherein: The method (200) is only applicable if the reference spectrum ( ) is performed in applications where multiple peak spectra are used, or included in each application, in the reference spectrum ( ) to distinguish between multi-peak spectra and smooth spectra; as well as In which the reference spectrum ( ) is at least one or each application of a multimodal spectrum, the error function (E(s)) is given by: the first item (T1); The first term (T1) and the negative spectral intensity value ( ) is the sum or weighted sum of the second term (T2) corresponding to the error of the difference spectral function (G) caused by The first term (T1) and the penalty factor (pen) are related to the negative spectral intensity values ​​( ) is the sum of the products of the second terms (T2) corresponding to the errors in the difference spectral function (G) caused by .

10. The method (200) according to any one of claims 1 to 4, further comprising, in each application: In the reference spectrum ( ) to distinguish between multi-peak spectra and smooth spectra; as well as Based on the distinction, the coefficient value is determined ( ), so that when the reference spectrum ( ) is a multi-peak spectrum, the coefficient value ( ) minimizes the error function (E(s)) and determines the coefficient value ( ), so that when the reference spectrum ( ) is a smooth spectrum, the coefficient value ( ) minimizes the replacement error function (RE(s)), where the replacement error function (RE(s)): is configured such that the reference spectrum ( ) is applied for the smoothed reference spectrum in comparison with the corresponding measured spectrum ( ) corresponds to the coefficient value ( ) shows an obvious minimum value; is given by the variance of said second derivative (SDG) of said difference spectrum function (G); or is given by the sum of the product of a first term corresponding to the corresponding measured spectrum ( ) of the spectral intensity values ​​( ) and the scaled reference spectrum ( ), the second term corresponds to the asymmetrically weighted sum of the squares of the differences between ) at the reference spectrum ( ) of the second-order derivative ( ), where , is applied to the corresponding measured spectrum ( ) of the spectral intensity values ​​( ) exceeds the reference spectrum ( ) of the corresponding spectral intensity value ( ) and the weighting factor of the square of the difference ( ) is smaller than that assigned to the corresponding measured spectrum due to ( ) of the spectral intensity values ​​( ) is smaller than the reference spectrum ( ) of the corresponding spectral intensity value ( ) and the weighting factor of the square of the difference ( ).

11. The method according to any one of claims 1 to 4, further comprising, in at least one or each application: Determine the reference spectrum ( ) whether it is a multi-peak spectrum; or By performing the reference spectroscopy ( ) by visual inspection to determine the reference spectrum ( ) is a multi-peak spectrum, when the reference spectrum ( ) includes at least one obvious peak, determine the reference spectrum ( ) is a multi-peak spectrum; or The reference spectrum is determined by: ) is a multi-peak spectrum: Determine the reference spectrum ( ) of the baseline ( ); Determine the reference spectrum ( ) and the baseline ( ) corresponds to the residual spectrum (R); determining, based on the residual spectrum (R), an index (KP) corresponding to the degree of multimodality of the residual spectrum (R); and Do at least one of the following: When the indicator (KP) is greater than a predetermined threshold, determining the reference spectrum ( ) is a multi-peak spectrum; as well as When the indicator (KP) is less than the predetermined threshold, determining the reference spectrum ( ) is a smoothed spectrum.

12. The method (200) according to claim 11, wherein: The indicator (KP) is determined by: determining the spectral intensity value ( ), or by determining the spectral intensity value ( ) divided by the spectral intensity value ( ) of the average value (RA); or Determine the variance of the residual spectrum (R).

13. The method (200) according to any one of claims 1 to 4, further comprising, based on the at least one or each Raman difference spectrum ( ), determining a measurement result (MR) of the at least one measured variable of the medium.

14. The method (200) according to any one of claims 1 to 4, further comprising: For the at least one measured variable, based on the method (200) based on a measured spectrum ( ) determined by the Raman difference spectrum ( ), determine the model (MOD) used to determine the measurement result (MR) of the corresponding measured variable.

15. The method (200) according to any one of claims 1 to 4, wherein: In at least one or each application, the interfering influences include at least one of: light pollution, luminescent and / or fluorescent light exhibiting at least one intensity peak, influences interfering with the analyte, the composition of the analyte and / or solvents contained in the medium, and background influences and / or broadband luminescent or fluorescent light exhibiting a smooth intensity profile; and / or Provide the reference spectrum ( ) includes at least one of the following: By means of a Raman spectrometer, at least one intensity spectrum of the interference effect is determined and provided, and based on these intensity spectra, the reference spectrum is determined and provided ( );as well as The reference spectrum is determined based on an intensity spectrum of at least one interfering analyte, a composition of the interfering analyte and / or a solvent contained in the medium. ).

16. The method according to any one of claims 1 to 4, further comprising, using the Raman spectrometer (110), repeatedly determining and providing a measurement spectrum ( ), and for at least one or each measured spectrum ( ) to determine and provide the corresponding Raman difference spectrum ( ),in: For each measured spectrum ( ), determine the Raman difference spectrum ( ) includes: based on the corresponding measured spectrum ( ) and the reference spectrum ( ), determine the corresponding measured spectrum ( ) corresponds to the coefficient value ( ), and based on the coefficient value ( ) to determine the Raman difference spectrum ( ); Based on the previously determined measured spectra ( ) determines the coefficient value ( ) is used to determine a series of subsequently determined measured spectra ( ) of the Raman difference spectrum ( );or Based on the newly determined measured spectrum ( ), by repeating the coefficient value previously performed ( ), at least once, repeatedly or regularly updating the continuously determined measured spectrum ( ) of the Raman difference spectrum ( ) of the coefficient value ( ).

17. A Raman spectroscopy measurement system (100), the Raman spectroscopy measurement system (100) being configured to perform the method (200) according to any one of claims 1 to 16, the Raman spectroscopy measurement system (100) comprising: A Raman spectrometer (110) configured to determine and provide a measured spectrum of a medium ( );as well as A difference spectrum determination unit (120), wherein the difference spectrum determination unit (120) is configured to determine the difference spectrum based on the measured spectrum ( ) and the measured spectrum ( ) contains the contribution of the interference effect to the reference spectrum ( ) to determine and provide Raman difference spectra ( ).

18. A non-transitory computer readable medium storing instructions which, when executed by one or more programmable processors, cause the one or more programmable processors to perform operations for performing a method (200) for Raman spectroscopy, the method comprising determining and providing a Raman difference spectrum ( ) for performing Raman spectroscopic measurements of at least one measured variable of a medium in at least one application, wherein The measured spectrum of the medium ( ) comprises at least partially obscuring the contribution of interfering influences of the spectral signature associated with the at least one measured variable, based on at least one measured spectrum of the medium determined by a Raman spectrometer (110) ) and the corresponding measured spectrum ( ) contains the contribution of the interference effect to the reference spectrum ( ), the operations include: Determine the spectral line ( ), the difference spectrum function (G) and the corresponding measured spectrum ( ) and scaling factor (s) with the reference spectrum ( ) corresponds to the difference between the products of Determine the corresponding measured spectrum ( ) corresponds to the coefficient value ( ), so that the coefficient value ( ) minimizes an error function (E(s)) representing the error of the difference spectrum function (G) as a function of the scaling factor (s), wherein the error function (E(s)) comprises a first term (T1) corresponding to the difference spectrum function (G) due to the reference spectrum ( ) and the error exhibited by the difference spectrum function (G), the first term (T1) includes at least one of the following: a first sum (S1) of the areas enclosed below the second-order derivative (SDG) of the difference spectral function (G) in all spectral regions where the second-order derivative (SDG) of the difference spectral function (G) is positive; and a second sum (S2) of the areas enclosed below the second derivative (SDG) of the difference spectral function (G) in all spectral regions where the second derivative (SDG) of the difference spectral function (G) is negative; and For at least one measurement spectrum ( ), determine and provide the corresponding measured spectrum ( ) and the coefficient value ( ) and the reference spectrum ( ) corresponds to the Raman difference spectrum ( ).

19. The Raman spectroscopy measurement system (200) according to claim 17, wherein: The Raman spectrometer (110) comprises: A monochromatic light source (1) is configured to emit a light having a predetermined excitation wavelength ( ) is emitted to a measurement area (3) of a sample (SM) configured to accommodate a medium to illuminate the sample (SM); and A spectrometry unit (7) configured to receive Raman scattered light (LR) emitted from an illuminated sample (SM) and to provide a measurement spectrum ( ), including an intensity spectrum of the Raman scattered light, wherein the spectrometry unit (7) includes a disperser (9), a detector (11) and a signal processor (13), and the signal processor (13) is configured to determine and provide the measurement spectrum based on a signal from the detector ( ) of the spectral intensity values ​​( );as well as The signal processor (13) of the spectrometric unit (7) and the difference spectrum determination unit (120) are cooperatively configured to perform the method (200) according to any one of claims 1 to 16.

20. The Raman spectrum measurement system (100) according to claim 17 or 19, further comprising a signal processing unit (130), wherein the signal processing unit (130) is configured to determine the Raman difference spectrum ( ) and based on the Raman difference spectrum ( ) for determining and providing the measurement result (MR) based on a previously determined model (MOD) for determining a measurement result (MR) of the at least one measured variable of the medium on the basis of spectral intensity values ​​of the measured variable.

Images